大型多輸入多輸出天線系統之相位雜訊抑制

Abstract

在無線通訊中，基地台具有大量天線之大型多輸入多輸出系統可以提供非常好的頻譜與能量的使用效率，且其使用效率可透過簡單的線性偵測與線性預先編碼來達到。因此，大型多輸入多輸出系統是有機會用於下一代無線通訊系統(5G) 與綠能通訊之關鍵的受矚目技術。然而，由於振盪器不理想特性產生之相位雜訊卻會造成無線通訊系統嚴重的效能損失。對於大型多輸入多輸出系統在上鏈與下鏈傳輸中，要解決相位雜訊的問題是困難的，因為它是一個多變數聯合相位雜訊估測與資料偵測問題。此問題之最佳解是難以推導的，且最佳解可能需要非常大的運算複雜度。在本論文中，我們研究解決大型多輸入多輸出系統之相位雜訊問題。首先，我們提出了一個低運算複雜度之相位雜訊抑制的方法於大型多輸入多輸出之上鏈系統。利用理想無相位雜訊損失之輸出信號雜訊比做為效能之上界，我們提出的上鏈方法，可以達到此無損失之上界效能。第二，我們也提出了一個低運算複雜度之預先編碼與相位雜訊抑制的方法於大型多輸入多輸出之下鏈系統。我們的研究也顯示，此提出的下鏈方法可以達到無相位雜訊損失之效能。因此，我們提出的上鏈與下鏈方法，提供了一個完整的基頻收發機架構於受到相位雜訊影響之大型多輸入多輸出系統，且其整體的運算複雜度也是相當低的。第三，基於所提出的上鏈與下鏈方法，我們設計了一個新的訊框(frame) 架構於頻分雙工(FDD) 與時分雙工(TDD) 兩種模式，兩種模式之效能比較也提供在文中，且此效能比較有助於決定在大型多輸入多輸出系統中應該使用何種模式。第四，我們考慮了單載波配合頻域等化器(SC-FDE) 傳輸於寬頻的大型多輸入多輸出之上鏈系統，並提供了一個低複雜度的方法來解決相位雜訊之問題。不同於正交分頻多工(OFDM) 系統，其中相位雜訊會造成共同相位誤差(CPE) 與載波間干擾(ICI)，我們的研究顯示，利用單載波配合頻域等化器的寬頻傳輸與我們所提出的相位雜訊抑制方法可以於寬頻大型多輸入多輸出之上鏈系統獲得無相位雜訊損失之效能。另一方面，在附錄E中，我們也提出了一個新的基於符號之接近最大概似偵測方法於多輸入多輸出天線(MIMO) 系統。不同於大型多輸入多輸出天線系統，在多輸入多輸出天線系統中，如果傳送天線數目與接收天線數目接近時，即使沒有相位雜訊，相較於最大概似偵測方法之效能，線性偵測方法具有一定程度之效能損失。我們提出的新方法不僅能達到接近最大概似方法之效能，還能提供資料不同之錯誤保護(UEP) 能力，也就是傳送於不同天線之資料，可以獲得不同之可靠度。

In wireless communications, massive multiple-input multiple-output (MIMO) systems in which the base station (BS) is equipped with a large number of antennas can provide significant spectral and energy efficiency by means of simple linear detection and linear precoding. Therefore, massive MIMO is an attractive technology for next generation (5G) wireless communication systems and for green communications. However, phase noise (PN) introduced by the impairment of oscillators can cause a severe performance loss in wireless communication systems. Solving the PN problem for massive MIMO systems in both the downlink and the uplink is challenging as it is a multivariate joint PN estimation and data detection problem. The optimal solution is difficult to derive and may lead to high complexity. In this dissertation, we focus on solving the PN problem for massive MIMO systems. First, we propose a low-complexity PN suppression scheme for massive MIMO uplink systems. By using the ideal PN free output signal-to-noise ratio (SNR) as a performance upper bound, we show that the proposed uplink scheme can achieve the upper bound. Second, we propose a low-complexity precoding scheme with PN suppression for massive MIMO downlink systems. We also show that the proposed downlink scheme has the potential to achieve PN free performance. Accordingly, the proposed uplink and downlink schemes provide a whole baseband transceiver structure of massive MIMO systems subject to the PN. We also show that the corresponding overall complexity of the proposed schemes is low. Third, based on the proposed uplink and downlink schemes, we recommend new and efficient frame structures for both the time-division duplex (TDD) and frequency-division duplex (FDD) modes. The corresponding performance comparison of the two modes is also provided, and the comparison may help determine which mode should be used for massive MIMO systems. Fourth, we consider single-carrier (SC) transmission with frequency-domain equalizer (FDE) for wideband massive MIMO uplink systems and propose a low-complexity PN suppression scheme to solve the PN problem. Unlike the orthogonal frequency-division multiplexing (OFDM), which suffers both the common phase error (CPE) and the inter carrier interference (ICI), we show that the SC-FDE has the potential to achieve PN free performance for wideband massive MIMO uplink systems. On the other hand, we propose a novel symbol-based near maximum likelihood (ML) detection scheme for MIMO systems in Appendix E. Unlink the massive MIMO systems, for MIMO systems, if the transmit and receive antenna numbers are close, the linear detection suffers considerable performance loss compared to the optimal ML detection even in the absence of PN. The proposed MIMO detection scheme not only provides the near ML performance but also the capability of unequal error protection (UEP), i.e., data from different transmit antennas can have different reliability.